MEASUREMENTS IN LIQUID FUEL SPRAYS

NASA Cooperative Agreement NCC 3-37

f (NASft-CE-177088) MEASUEEMEN1S IN LIQUID N86-/3Q960.--FUEL SPR&YS Final fieport, 15 Dec. ,1983 - 14 .Dec.,,1985 JCarnegie-Mellon Ociv.), 51 p .. .

CSCL 20D OnclasG3/34 43377

FINAL REPORT

12/15/83 - 12/14/85

Norman ChigierChien-Pei Mao

Department of Mechanical EngineeringCarnegie-Mellon University

Pittsburgh, PA I52I3

https://ntrs.nasa.gov/search.jsp?R=19860021488 2020-04-28T23:47:21+00:00Z

1. INTRODUCTION

To successfully achieve all weather operation of military aircraft and rotorcraft,

a thorough understanding of the adverse weather effects on the aircraft performance

is required. It is for this reason that research on the effects of icing, snow and

heavy rain has been carried out at NASA for several years. So far, most of the tests

have been performed using expensive, time consuming and dangerous flight testing. A

unique ground test facility is being established at NASA Lewis Research Center to

simulate the environmental and flight conditions needed to study adverse weather

effects. One of the most important components in this ground test facility is the

water spray system which consists of many nozzles fitted on spray bars. Water is

injected through air-assisted atomizers to generate uniform size drops to simulate

natural icing in clouds.

The air-assist atomizer uses pressurized air to enhance the atomization produced

by pressurized liquid. The atomization of liquid is primarily achieved by the shear

forces due to high speed air flows. The major requirement is to generate uniform

size water drops with control at the test section of the wind tunnel so that realistic

environments can be achieved. Typical water drop sizes in icing clouds range from 5

to 60 //m, mass median diameter. The performance and characteristics of the air-

assisted atomizer designed by NASA Lewis are being studied at CMU in order to

determine the requirements for simulating icing clouds. Therefore, the primary

objective of the present investigation is to provide experimental data on drop size

distribution over a wide range of operating conditions. Correlation equations for mean

drop size and initial injection parameters are being determined to assist in the design

and modification of the Altitude Wind Tunnel. Special emphasis is being placed on

the study of the aerodynamic structure of the air-assisted atomizer sprays. Detailed

measurements of the variation of drop size distribution and velocity as a function of

time and space are being made. A second objective is to provide accurate initial

and boundary conditions for computer model evaluation.

A number of instruments including high magnification photography, laser

diffraction, and the phase doppler analyzer are being used to make the measurements

of spray angle, line-of-sight drop size distribution, and local point information such

as drop size, velocity, liquid flux, and number density. The current drop velocity

measurements using the phase doppler analyzer are limited to low liquid injection

conditions due to the restricted instrument response. The higher frequency signal

measurements will be pursued by adding optical frequency shifting such that the

Doppler frequency of the scattered light signal is optically downshifted into the

operating range of the processor.

A comparative study of the data being obtained by photography, laser

diffraction, and phase doppler on mean drop size distribution has shown inconsistent

results at the same operating flow conditions. The differences in SMD are related to

the limitations in dynamic range, detector response, sampling techniques, and

sampling time of each instrument. An "ultimate" SMD could be measured provided

that these limitations are suppressed and all drops are measured over a "long

enough" time to cover all the possible low and high frequencies in the systems.

None of the instruments are currently capable of measuring this "ultimate" SMD.

Therefore, comparative measurements using different instruments are always subject

to a certain degree of disagreement. Photography has the problems of obtaining

statistically representative results and of not counting extremely large numbers of

unfocused small drops. The laser diffraction particle sizer is limited by the threshold

sensitivity of the diodes and the small detecting area of the inner diode rings. The

relatively small number of large drops are overshadowed by a huge amount of small

drops which reduce the sensitivity of the inner diodes which have relatively small

area. In principle, the phase doppler particle analyzer has the potential of detecting

the "ultimate" SMD of the sprays. It also provides the capability of examining

particular sections of the size distribution spectrum that are of particular interest, by

varying the threshold level of the processor and the high voltage of the detector.

However, this capability is restricted by the instrument dynamic range and detector

response.

2. EXPERIMENTS

The aim of the present investigation is to find correlation equations for

predicting SMD (Sauter Mean Diameter) over a wide range of injection conditions to

assist icing tunnel design and to establish a data base for computer model

evaluation. The water mass flow rates that are being studied range from 10 Ibm/hr to

100 Ibm/hr. This corresponds to injection pressures ranging from 10 psi to 205 psi.

The mass flow rate of air ranges from 20 Ibm/hr to 150 Ibm/hr which corresponds to

a pressure injection range from 10 to 155 psi. Experiments are carried out in two

different sizes of spray chambers under well-controlled flow conditions to

accommodate both high and low water/air mass flow rate ratios.

Spray Chambers

Two spray chambers are being used for testing. Fig. 1 shows the configuration

of the small scale spray chamber. This chamber was designed to accommodate

maximum water flow rates of 40 Ibm/hr. The cross-sectional area of the chamber is

12x12 inches and the window viewing area is 12x14 inches. A traversing system

installed in the chamber provides 2-D freedom of motion of the atomizer without

requiring the diagnostic instruments to be moved. For injection conditions of water/air

mass flow rate ratio less than 1, strong recirculating flows are created in the

chamber which significantly perturb the sprays. The large scale spray chamber is,

therefore, used for examining the high speed liquid flows. Fig. 2 shows a schematic

of the large scale spray chamber. This system is capable of providing maximum flow

rates of 1200 Ibm/hr for water and 190 Ibm/hr for air. The four-sided, glass walled

spray chamber has a cross sectional area of 20x25 inches. The window height is 29

inches. «A secondary air flow, produced by the suction of a high power blower,

eliminates the flow recirculation and fogging of the window. The advantages of the

large scale chamber are: wider range of test conditions, no recirculation,

measurements can be made farther downstream (up to 30 inches downstream from

the atomizer orifice), and a larger working distance. The disadvantage is that the

larger cross sectional distance is approaching the critical value of vignetting so that

receiving lenses less than 300 mm cannot be used for data acquisition. The small

spray chamber, on the contrary, is much easier to operate and allows more accurate

control of the flow rates.

Atomizers

Two atomizers have been received from NASA for testing. They are numbered

as nozzle 100 and 2, respectively. These two atomizers have the same nozzle design

(see Fig. 3), but with different flow rates at the same injection conditions. A careful

calibration of the water volume flow rates of these two atomizers was made. Fig. 4

shows the water volume flow rates versus differential pressures for both the number

100 and 2 atomizers. The "differential pressure" refers to the pressure difference

between water injection and air injection at the atomizer exit in units of psi. Both the

number 100 and number 2 atomizers have been tested in the present investigation.

Qualitatively, these two atomizers have similar characteristics. As indicated in Fig. 4,

#100 nozzle has a higher volume flow rate than that of #2 nozzle at the same

differential pressures. For nozzle #100, all data of volume flow rates fit well with

the following equation,

v = 0.81

where v is the water volume flow rate in gallons per hour, and 5P is the differential

pressure in psi. As for nozzle #2, it was also found that water volume flow rate is

proportional to the square root of differential pressure,

v = 0.609

Although these two nozzles have the same design, the water mass flow rates are

different. The differences are found due to the surface roughness at the nozzle exit.

Nozzle #100 has much rougher lips at the exit. The nominal diameter of the orifice

is essentially bigger for nozzle 100. In the #100 air-assisted atomizer, the pressure

of air flow affects the pressure of liquid injection. Fig. 5 shows the mass flow rate

of air versus gauge pressure with and without water injection for nozzle #100. The

mass flow rates of air are linearly proportional to the gauge pressure in the nozzle.

Under the water injection condition, the mass flow rates of air are slightly decreased

due to the decrease of the cross-sectional area at the nozzle exit. Fig. 6 shows the

air mass flow rates versus the gauge pressure for nozzle #2.

Photography

Still photography is used as the standard means for visualization and

preassessment of global spray structure. It also provides a calibration intermediate

between the standard calibrating devices (reticule and impulsed drop generator) and

other instruments with high data acquisition rate. A low magnification 35 mm Canon

F-1 camera with near forward lighting technique is used to determine the spray angle.

The light source is a Koolbeam light bulb array which consists of 300 W tungsten

halogen lamps. Measurements of the spray angle are made by setting the camera at

f/11 and 1/125 second time exposure. These settings are selected to provide high

contrast and consistent spray angles since spray angle can be changed by varying the

light exposure. A high magnification 4x5 inches format camera with a 5 feet and a 10

feet long bellows is also used to examine local spray structure. The high speed, high

intensity EG&G 549 microflash light source is utilized to freeze the high speed drop

motion and to provide enough light scattering intensity from the smaller drops.

Laser Diffraction Particle Sizer

For a rapid analysis of the spray characteristics, the laser diffraction particle

sizer is currently the most effective instrument at CMU. Line-of-sight mean drop size,

size distribution and liquid volume concentration information are provided by this

instrument. The instrument is based on the detection of near-forward Fraunhofer

diffraction. It is rather simple in operation and is efficient in data acquisition.

However, there are many limitations associated with the detection technique.

Extensive studies of the accuracy and limitations of the laser diffraction particle

sizer have been made at CMU. The major problem with this instrument is obtaining

accurate drop size spectra in sprays due to the lack of the capability of "seeing" the

small fraction of large drops which are clearly seen in photographic pictures. Results

have shown that this problem is more severe at extremely high and low obscuration

conditions. Nevertheless, the laser diffraction particle sizer is used as the principal

instrument for determining representative drop sizes in sprays. Measurements are

made at several selected downstream axial and radial stations. At each axial location,

the atomizer is traversed radially across the laser beam to obtain radial size

distributions. A horizontal plane averaged SMD is obtained by taking into account the

non-uniform liquid volume concentration distribution for drop size correlation.

Phase Doppler Spray Analyzer

This instrument is used for detailed local drop size and velocity measurements

in the sprays. It was developed by Bachalo and Houser1. The technique is based upon

the measurement of the interference fringe pattern produced by spherical drops

passing through the intersection of two laser beams. Three detectors, separated at

fixed spacing, are used to receive doppler signals and to determine the phase shift

due to different path lengths of the laser beam. The spatial frequency of the

interference fringe pattern is linearly proportional to the measured drop size. The

most important capability of this instrument is that it provides simultaneous drop

size and velocity measurements. Other local point information provided by this

instrument include: liquid flux, number density, size-velocity correlation, spatial and

temporal mean drop sizes. A detailed validation study has been performed to

determine the limitations and accuracy of this instrument. A rotating disk and an

impulsed piezoelectric drop generator are used as calibration devices. A number of

problems with this instrument have been found during the validation study. They

include: an inconsistency in measured velocities which differed at times by a factor

of 2, a complete lock-up at certain filter selections, and general failure of certain

configurations to provide reliable data. These problems have been corrected by the

manufacturer and they were recognized to be caused by a digital board trace short,

incomplete trace cuts on the processor motherboard, and improper plastic connector

shells on cables. From our experience, all optical configurations (which determine

different measurement ranges) require to be examined before use. Otherwise, any of

the problems mentioned above may arise with certain optical configurations and may

not be easily detected. Our instrument has an upper doppler frequency limit of 3.2

MHz due to the limitation of the detector/preamplifier response. It is for this reason

that the phase doppler spray analyzer is currently being used only for measuring very

low water injection conditions which have lower drop velocities. The system will be

upgraded by purchasing a rotating grating frequency shifter to enable higher speed

flows to be measured so that the whole range of injection conditions can be

examined.

3. RESULTS AND DISCUSSION

Breakup Mechanism and Atomization

Fig. 7 to Fig. 14 are shown to demonstrate the typical air-assist spray structure

and the process of atomization. Air-assist atomizers employ the kinetic energy of

air flow to tear the water jet into ligaments and subsequently into various sizes of

drops. Fig. 7 shows that water emerges from the orifice as a circular jet with very

high speed. The water circular jet is very unstable, even under the condition of no

air injection. The oscillation of the jet causes the formation of necking and the

amplitude of the oscillation grows with downstream distance. Any distortion of the

water jet surface due to oscillation is opposed by surface tension forces. Through

several stages of force balance, ligaments and satellite drops are formed between

the necks. The atomization can be highly improved by injecting the water into a

high-velocity airstream. Fig. 8 to Fig. 14 show the effect of increasing air velocity

on atomization. Apparently, Fig. 14 shows the best atomization due to higher air

velocity. In fact, the NASA air-assist atomizer is very similar to the

Nukiyama-Tanasawa2 atomizer. At a fixed water injection pressure, the mean drop

size decreases with the increase of air flow rates or decreases with the increase of

relative velocity. In the flow field near to the atomizer, the velocity of air is much

higher than that of the water. The velocity of air was estimated to be close to

sonic velocity at all test conditions. The inverted "V" shape of the spray near the"

orifice indicates that the water jet is surrounded by high speed air flow and is

dragged by the air in the direction of the air flow. The velocity of air relaxes much

faster than the liquid flow velocity as downstream distance increases. The global

shape of the spray is seen to be converted into a "V" shape far downstream where

the velocity of liquid is faster than the velocity of air. It is also noted that the

angle of the inverted "V" shape in the near flow field increases with the increase of

air mass flow rates. Thus, the increase of air velocity increases the surface of

interaction between the liquid and air which enhances the atomization. In the very

inner section of the water jet, a group of ligaments and large drops retain their

initial high velocity without too much interaction with the surrounding air. The

penetration distance of ligaments and large drops, in the cases of fixed liquid

injection pressure, depends on the initial air mass flow rates. Since the relative

velocity between water and air is kept the same for Figs. 8 to 14, the mass flow

rate of air is the only dominant factor that influences the quality of atomization.

More than 200 photographic observations show that the NASA air-assist atomizer

spray structure is very similar to that of diesel jet sprays which are being studied

separately at Carnegie Mellon University. The breakup processes occur periodically

and consist of a series of inverted "V" shape sprays. The liquid water is mostly

concentrated in the center of each inverted "V" group and becomes more dilute as

the "V" shape spreads out radially. Scattered satellite drops were found between

each inverted "V" group. This observation shows that the spray breakup is not a

continuous process but very oscillating. Measurements should be made with "long

enough" time so that a statistically representative result can be assured.

Figs. 15 to 20 show the close-up look of NASA air-assist atomizer sprays at

several centerline downstream locations at the condition of water injection 58.7 psi

and air injection 10 psi. The use of the 4x5 inch format camera with a 10 ft.

bellows allows us to visualize an area of 9x9 mm2 inside the sprays. These pictures

show much more detail about the breakup process and drop size distribution. Fig. 15

shows that the initial atomization starts at the periphery of the water jet. The shear

forces tear and shatter the water jet surface apart. The center core is so dense that

almost no light can penetrate. Fig. 17 shows that long ligaments are formed at x=4.5

cm downstream. The surface tension is struggling to rearrange the shape of the

ligaments. However, adverse pressure of air tends to further break down. The

process of atomization is actually slowed down with downstream distance because

of the quick deceleration of the air flows. The redistribution of drop size and drop

velocity at downstream stations causes collision between drops. In Fig. 21, a few

large drops are about to separate from the small drops after collision. The bridge

between them is the source of satellite drops after the breakup. The phenomenon of

drop collision has been studied by several researchers3'4 in the past. The effect of

collision on drop size distribution can be categorized into two groups: (1) drops

coalesce after collision , and (2) satellite drops are created after collision. The

former effect leads to an increase in the mean drop size and the latter effect leads

to a decrease in the mean drop size. The efficiency of collision depends on the

Weber number at the moment of collision. A two color phase doppler system

designed for measuring two components of drop velocity can be used to evaluate

the efficiency of collision at a local point by calculating the Weber number of size

bins. By observing these magnified pictures, a small number of large drops in the

size range from 300 //m to 500 pm diameter can be easily seen even at downstream

distance 13.5 cm where atomization is already completed. Comparisons between

photographs and data obtained from the laser diffraction particle sizer have shown

inconsistencies in drop size distribution, especially regarding the large drops in the

sprays. Further analysis of the photographic results using an automated digital

imaging system5 is necessary to provide long time statistical averaging drop size

information.

The spray cone angle is also measured using a near forward lighting technique.

All pictures were taken under the same optical setting and with the same films. The

camera was set at the f/11 aperture with a 1/125 second time exposure. A Kodak

Tri-X film was used to obtain high contrast pictures. The spray angle was measured

over a wide range of operating conditions. The water injection pressure was tested

from 35 psi to 325 psi and the air injection pressure was varied from 25 psi to 155

psi. There was no special trend found for the spray angle with respect to the

pressure variation. It actually oscillates with pressure. In general, the global spray

angle varies between 19° and 25°. The fine drops that surround the main water jet

affect the accurate determination of the spray boundary due to longer exposure time.

However, the angle of the inverted "V" shape in the flow field near the atomizer

varies with the mass flow rates of air (see Fig. 8 to 14). The higher the mass flow

rate of air, the wider the angle of the inverted "V" shape. Fig. 23 gives an example

of the picture obtained using the near forward lighting technique for the

determination of spray cone angle.

Downstream Drop Size Variation

Line-of-sight mean drop size measurements were first made using nozzle #100

at various downstream locations over a wide range of injection conditions. Since

the distance for complete breakup was quite long, the 4x5 format camera was used

to examine the first downstream measurement station. Fig. 24 shows the radial SMD

distribution at several downstream stations at P = 35 psi and P . = 25 psi for nozzlew air r

#100. The size distribution is found to be quite uniform across the sprays at all

locations. But, the mean drop sizes increase with the increase of downstream

distance. Fig. 25 shows the radial SMD distribution at several downstream stations

for P = 65 psi and P . = 25 psi. By comparing Fig. 25 with Fig. 24, the radial size

distribution is no longer uniform across the spray due to the increase of water

injection pressure. The maximum SMD peaks at a position slightly off the center

axis. A common feature for both cases is that the SMD increases gradually with the

downstream distance. Further increase of the water injection pressure increases the

non-uniformity of the radial size distribution. The symmetry of the size distribution

is also grsatly distorted at higher water injection conditions. Fig. 26 indicates the

shifting of the peak SMD position as a function of distance at P = 200 psi and P . =

25 psi. As can be seen, the mean drop size varies considerably across the sprays.

There is no single SMD that can be used to represent the entire spray. A judicious

decision has to be made for selecting appropriate values of SMD to fit existing

empirical correlation equations. Further details of drop size correlation will be given

in the next section.

In order to simulate the icing clouds in the wind tunnel, it is required that a

uniform drop size distribution be produced by the nozzle. Based on the

measurements made at many axial and radial stations over a wide range of injection

conditions on nozzle #100, it was found that the structure of NASA air-assist

atomizer sprays can be separated into two categories depending on the water/air

mass flow rate ratios. For a water/air mass flow rate ratio of less than 1, no

matter how high the injection pressures were, the radial mean drop size distribution

is rather uniform as seen in Fig. 24 and gradually increases with axial downstream

distance. For water/air mass flow rate ratio greater than 1, the radial mean drop size

distribution is nonuniform. It peaks at a position somewhere off the center axis.

The deviation of the peak SMD position from the center line increases with the

downstream distance. The characteristic of the NASA air-assist atomizer sprays of

not being symmetric as water/air mass flow rate ratios increases is not completely

due to the aerodynamic instability effect. It is partially due to the original tilting of

the water jet tube inside the nozzle. Nevertheless, the nonsymmetry of the spray is

not going to affect the evaluation and correlation of size distribution based upon

planar averaged drop size distribution at each axial station. Fig. 27 shows another

example of a uniform radial SMD distribution at high water injection pressure

conditions for nozzle #100. The water/air mass flow rate ratio is close to 1. Fig. 28

indicates that the mean drop size is increasing along the downstream distance and

the size distribution is not very uniform across the sprays for water/air mass flow

rate ratios much greater than 1. For the purpose of simulating icing clouds, it is

therefore suggested that the operation of nozzle injection pressures be maintained at

the conditions of water/air mass flow rate ratios less than 1 so that uniform drop

size can be produced. Up to this point, all discussions have been based upon the

results obtained for NASA nozzle #100. From this point on, all of the tests were

conducted using NASA nozzle #2, and measurements were made for the injection

conditions that provide a water/air mass flow rate ratio less than 1. Nozzle #2 was

10

manufactured to high precision, and it is identical to other nozzles (Ref. 6) also being

tested by NASA Lewis Research Center.

In general, these two nozzles (100 and 2) showed similar characteristics in spray

structures. The size distribution, however, differs in magnitude. Further details about

the NASA air-assist atomizer sprays are discussed based on the measurements of

nozzle #2. Fig. 29 shows a picture taken by the high magnification 4x5 inches format

camera at 15 cm downstream from the atomizer with water injection pressure 59 psi

and air injection pressure 10 psi. By comparing many photographs such as this and

laser diffraction measurements at the same locations, it has become clear that there

is a number of large drops in the sprays. This raises the question of whether the

uniform drop size distribution measured by the laser diffraction particle sizer is

accurate or not. These large drops that are seen in the photographs can play an

important role in the simulation of icing clouds in the altitude wind tunnel. In

addition, the reported phenomena of continuous increase in measured downstream

drop size has to be carefully examined to establish whether water drops continue to

increase in size until they reach the test section of the wind tunnel. Currently, the

effects of drop acceleration/deceleration, collision/coalescence, and evaporation on

downstream size variation are being studied in our laboratory. Evidence has been

found that drops do have a strong tendency to coalesce after collision3. The

deceleration of drops causes interaction and collision and results in a subsequent

increase in drop size.

Another problem associated with the laser diffraction technique is the

preselection of the size distribution mode. Throughout the present investigation, a

Rosin-Rammler size distribution function was chosen R = exp[- {D/x)N], where R is the

volume fraction of particles with a diameter greater than D, N is the parameter

related to the polydispersity of the distribution, and x is a representative mean

diameter such that 36.8% of the total particle volume is greater than x. However, it is

found that this size distribution function may not be valid throughout the sprays. Fig.

30 shows an example of a good fit using the Rosin-Rammler (2 parameters) and

model independent (15 parameters) modes to obtain the weight fraction distribution.

In the region inside the spray, with higher obscuration, the Rosin-Rammler mode is

acceptable for obtaining accurate size distribution. However, the Rosin-Rammler

mode fails to predict the large size drops at the edge of the sprays where there is

low obscuration. Fig. 31 shows that the model independent mode provides a more

accurate determination of the large drops that are present near the edge of the

11

sprays. In general, the Rosin-Rammler mode can be applied to most areas of the

NASA atomizer sprays except the region near the spray boundaries where the model

independent mode has to be used in order to get more accurate SMDs. However, it

was also noted that the Rosin-Rammler size distribution mode is not appropriate for

drop size calculations even at the center positions under extremely high water/air

mass flow rate ratios (>1).

Mean Drop Size Correlation

Empirical mean drop size correlation equations have been widely used for

describing the characteristics and structure of various types of atomizer sprays.

There is no universal equation that is suitable for all sprays. In the present study

for NASA air-assist atomizer sprays, drop size correlations as a function of initial

input parameters were determined based upon the line-of-sight laser diffraction

particle sizer measurements for nozzle #2. Since the drop size varies all over the

spray, it has been arbitrarily selected that all measurements be made at a fixed axial

location 25 inches downstream from the atomizer where complete atomization is

definitely achieved for all conditions. Fig. 32 shows the radial SMD distributions for

several different water injection pressures. The air injection pressure was kept the

same at 65 psi. All five test conditions were measured at the axial station 25

inches from the nozzle, and all water/air mass flow rates are less than 1. Although

all test conditions have uniform drop size distribution, a planar averaged SMD is

calculated from each local radial SMD across the spray by taking into account the

radial liquid volume concentration distribution. A plane averaged SMD is more

appropriate to represent the entire spray and to derive empirical correlation

equations. Fig. 33 is a summary of the variation of SMD with respect to differential

pressures (PW~P ir) for the NASA #2 nozzle. At fixed air injection pressures, the SMD

increases with the increase in water injection pressure. At fixed water injection

pressure the SMD decreases with the increase of air pressures. There is a more

significant increase in SMD by increasing water injection pressures at lower air

pressures. At the high air injection pressures, the increase of water pressure or flow

rates has no effect on mean drop size variation.

There are many published equations for the prediction of SMD of various types

of atomizers. Several data fit equations are considered based upon the

characteristics of the nozzle, physical process of atomization, and possible dominant

parameters of the injection fluid flows.

12

/ rn \o.s(1) SMD ~ V 1 *vt-/ /

m

(2) SMD ~ mL/mA

(3) SMD/ . . \I.B

~ \ mL/mA /

where VR is the initial relative velocity between the water and air. None of the

above correlations fit the data as well as the Simmons SF7 equation7. Simmons

derived an equation for air-assist simplex nozzles based upon the concept of

hypothetical film thickness and relative interaction of liquid and air. We found that

this equation also predicts very well the SMD for the NASA air-assist atomizer. Fig.

34 shows the results of the SMD correlation with respect to the grouping parameter,

-0.325 / m \ 0.55SMD ~ P \. L. ;mL VL + mAVA

*where />A is the density of air at injection pressure, ML is the mass flow rate•

of water, VL is the exit velocity of liquid, MA is the air mass flow rate, and VA is

the air exit velocity. This correlation equation can provide useful information about

drop size distribution at different operating conditions to assist in wind tunnel

design.

Local Drop Size and Velocity Measurements

The measurements of local drop size and velocity are extremely important for

spray structure analysis and computer model evaluation. The newly developed light

scattering technique, the phase doppler spray analyzer, is being used to provide

information on local mean velocity, velocity fluctuations, SMD, liquid flux, and

number density. Since the instrument has an upper doppler frequency limit of 3.2 MHz

and a dynamic range of 30, several injection conditions within the measurable range

were selected for tests in the small scale chamber. The maximum downstream

distance that can be measured is 17 cm from the nozzle exit. Table 1 is a summary

of the phase doppler spray analyzer results for water flow rate of 45.3 Ibm/hr and

air flow rate of 19.4 Ibm/hr at axial station 17 cm. All current test conditions are

limited to water/air mass flow ratios greater than 1 and they all show qualitatively

similar spray structures. In general, the spray structure for water/air mass flow rate

greater than 1 can be summarized as follows: the radial profile of mean axial drop

13

velocity sharply peaks at the center of the sprays. Most of the drops travelling in

the center line region possess very high speed which is independent of drop size.

There is no indication of mean drop velocity relaxation along the centerline within

the 17 cm distance that has been measured. The measured velocity fluctuations (rms)

have shown a gradual increase along the radial distance with a subsequent decrease

towards the outer edge of the spray. Liquid volume flux (cc/sec) is also determined

from the phase doppler spray analyzer measurement; these show almost the same

radial profiles and magnitude at various axial locations. This indicates that spray

evaporation is insignificant within the measured downstream distance. The liquid

volume flux peaks at the centerline and then decreases sharply along the radial

distance. The radial number density normally increases from the center axis and

reaches a maximum at a point close to the edge of the spray. The higher number

density region will result in higher possibilities of drop collision and coalescence.

The increasing effect of drop collision toward the edge of the spray results in an

increase of the drop size' spectrum. Photographic evidence indicates that collisions

are taking place. The radial drop size distribution indicates that the local SMD

decreases along the radial distance from a peak at the centerline of the sprays. By

comparing with the line-of-sight laser diffraction measurements, the local SMDs

measured by the phase doppler analyzer are much larger. Since the laser diffraction

particle sizer provides the particle size information in an in-line method with rather

poor spatial resolution, there is a small quantity of large drops which are not "seen"

by this instrument. The results reported by the laser diffraction technique can thus be

in error. The inconsistency in the drop size distributions reported by these two

techniques needs to be resolved by the imaging technique. The development of an

automated digital imaging system is underway in our laboratories to assist the

comparative study of spray structures using different instruments.

4. SUMMARY

In the present investigation, two well controlled spray facilities for testing

NASA air-assist atomizers used in icing research have been constructed and

calibrated. Interesting results about the spray aerodynamic structure have been

observed. Under all test conditions, it was found that water/air mass flow rate ratio

is a useful parameter to determine whether or not the drop sizes are uniform across

the sprays. For water/air mass flow rate ratio greater than 1, the radial mean drop

size is not uniform across the sprays. For water/air mass flow rate ratios less than

1, the radial mean drop size is uniformly distributed across the sprays. The gradual

increase of SMD along downstream distance is a phenomenum associated with the

aerodynamic effects and needs to be examined at distances farther downstream (>30

14

inches from atomizer). Drop deceleration and collision are the main factors causing

SMD to increase along downstream distance. Simmons7 drop size correlation

equation was found to be suitable for predicting the SMD's for NASA air-assist

atomizers over a wide range of injection conditions.

Comparisons between laser diffraction and photographic measurements are

leading us to believe that the small area of the inner diodes, the threshold sensitivity

of the diodes, and the relatively small number of large drops are resulting in

insensitivity of the laser diffraction technique to the presence of the larger drops.

For this reason, the laser diffraction instrument reports much smaller mean drop

sizes at various locations than those measured by photography. However, it also

needs to be pointed out that the photographic results are also neglecting a very large

quantity of small drops. True "ultimate" SMD is unable to be measured by any of

these two techniques.

A newly developed drop sizing instrument, phase doppler spray analyzer, was

successfully applied to measure high speed flows in NASA atomizer sprays. Useful

information such as drop velocities, liquid fluxes, drop number density, and SMD for

spray analysis was obtained for several low injection pressure conditions. Our

present processor for the phase doppler analyzer has an upper doppler frequency

limit which restricts the measurements to sprays with lower speed flows. For future

work, the phase doppler will be modified to use optical frequency shifting such that

the doppler frequency of the scattered light signals are optically downshifted within

the operating range of the processor. By applying frequency shifting to our present

processor, drop velocity up to 300 m/sec can be measured. A two color phase

doppler system is also under consideration for making two-component velocity

measurements simultaneously. With two-component velocity information, it is

possible to determine drop trajectories and to evaluate the effect of drop collision in

the sprays. We also intend to use the control of detector voltage, threshold, and

high and low pass filtering of the processor for closer examination of the larger

drops. Using the concept of a spectrometer with a narrow band width, we can

progressively tune in to sections of the size distribution spectrum and obtain more

information about certain size ranges with less interference by not processing

particles outside the size range of interest. In this way, we can make a more

meaningful comparative study between the photographic and phase doppler results.

15

REFERENCES

1. Bachalo, W.D. and Houser, M.J., "Phase Doppler Spray Analyzer forSimultaneous Measurements of Drop Size and Velocity Distributions,"Optical Engineering, Vol. 23, No. 5, pp. 583-590. 1984.

2. Nukiyama, S. and Tanasawa, Y., "Experiments on the Atomization ofLiquids in an Airstream", Trans. Soc. Mech. Eng. Japan, Vol. 5, pp. 68-75,I939.

3. Brazier,P.R. et. al, "The Interaction of Falling Water Drops: Coalescence,"Proc. R. Soc. Lond., A. 326. pp. 393-408, 1972.

4. Podvysotsky, A.M. and Shraiber, A.A., "Coalescence and Break-Up of Dropsin Two-Phase Flows", Int. J. Multiphase Flow, Vol. 10, No. 2, pp. 195-209.1984.

5. Ahlers, K.D. and Alexander, D.R.. "A Microcomputer-Based Digital ImageProcessing System Developed to Count and Size Laser-Generated SmallParticle Images," Optical Engineering, In Press, 1985.

6. Tacina, B., NASA Lewis Research Center, Private Communication, 1985.

7. Simmons, H.C., "The Prediction of Sauter Mean Diameter for Gas TurbineFuel Nozzles of Different Types," ASME Paper no. 79-WA/GT-5, 1979.

TABLE 1

A Summary of Phase/Doppler Results for Water injection 45.3 Ibm/hr and Air Injection 19.4 Ibm/hr

Left Hand Side Right Hand SideRadialPosition 2.5 2 1.5 1 0.5 0 0.5 1 1.5 2 2.5r(cm)

MeanVelocity 1.36 4.4 8.9 15.4 28.7 41.2 39.8 20.9 10.9 6.8 3.9(m/sec)

VelocityFluctuation 0.85 2.5 4.6 5.9 7.5 4.6 6.4 7.4 5.3 3.2 2.1(m/sec)

| • VOX * _Q fj _g g f _ ̂| f _ g g _^ M

Flux 7.6x10"° 5.9x10"' 3.3x10"° 9.0xlO~° 1.8xlO~D 1.8xlO~* 8.2xlO~3 7.9xlO~° 2.6xlO~° 1.3x10"° 3.0x10"'

(cc/sec)

NumberDensity 79 117 120 164 77 58 67 128 216 232 112(l/cc)

TemporalSMD 28 34 46 49 62 131 92 50 34 30 28(um)

Hater/Air Mass Flovrate Ratio: 2.34/ x = 17 cm

FLOWSTRAIGHTENE

ENTRY PORT

NOZZLE

WINDOW

EXIT PORT

SPRAY CHAMBER

Figure 1 Schematic of the small scale spray chamber.

4"x5" ViewCamera \ ' «

ViewingfWindows

SheetMetal

Secondary AirIntake Shroudand FlowStraighteners

Floor Level

—v\Air Exhaust \ \to Blowerer }

CatchTank

fr

Figure 2 Schematic of the large scale spray chamber.

\Y\i i.\\\\\

Air

Figure 3 Cross-sectional view of NASA air-assist atomizer.

WATER VOLUME FLOW RATE CALIBRATION

- NASA NOZZLE ftl (ref,6)

60 80 100 li0 S0DIFFERENTIAL PRESSURE (psid)

200

Figure 4 Water volume flow rates versus pressure difference between waterand air (P - P ) at atomizer exit.

w a

150AIR MASS FLOW RATE CALIBRATION CURVE

- NO WATER INJECTION

• WITH WATER INJECTION

AIR PRESSURE (psig)

Figure 5 Air mass flow rates versus air injection pressure for nozzle #100.

AIR MASS FLOW RATE CALIBRATION CURVE

- NO WATER INJECTION

AIR PRESSURE (psig)

Figure 6 Air mass flow rate versus air injection pressure for nozzle #2.

126

Figure 7

A.ir

T>S1,

Figure 8

T>e

F>air=16.O

Figure 9

26.6 T>©i

Figure 10

F>wetter—126 p>ei, Pair=36.O T>ei

Figure 11

126 T>e

Figure 12

F* wetter ==126 T>ei, F>etir=56.5 T>ei

Figure 13

F> wetter =126 T>ei, :Pctir=6T.8

Figure 14

.T T>ei, — 1O T>ei

cm

Figure 15

1O T>ei

cm

Figure 16

ORIGINAL PAGF pOF POOR QUALITY

£HNVV . .i?wv>;

13 wetter—=58. T ^ T>s

cm

Figure 17

I % . f-.. .*. ; -

,r—58.7 T>ei, I>»ir—1O T>ei

>L=6.3 cm

Figure 18

ORIGINAL PAG" rS

OF POOR QUALITY

«&•*; *• .• • •

£•-•'• +':*X-f**\ i{ • ' ; • • • • 1» •• f ^

v> t • • ' . - •*•*•.-*.•

T ¥>Bi,

X=8.O cm

Figure 19

—58.7 T>si, Pair—1O T>si

1O.O cm

Figure 20

ORIGINAL PAGE ISOF POOR QUALITY

•

T T>ei, — 1O T>ei

11.T cm

Figure 21

:Pwetter—58.T T>ei, Fair—1O T>si

X=13.5 cm

Figure 22

ORIGINAL TAGS BOF POOR QUAU7Y

Figure 23 Spray photograph obtained by using near forward lighting techniquefor the determination of spray cone angle.

SHD RADIAL DISTRIBUTION

X=43 CM• X-58 CN

•4 -3 -2 -1RADIAL DISTANCE (en)

pi-35 psi. R.-25 psi

Figure 24 Radial SMD distribution at various downstream stations at P = 35psi and Psjr= 25 psi for nozzle #100.

SMD RADIAL DISTRIBUTION

19-18171615:1

* X=28 CND X=38 CNo X:48 CN

X=33 CNX=43 CN

•3 -2 -i 4 rRADIAL DISTANCE (CN)

l» =*6S psi, p.-25 p:>i

4

Figure 25 Radial SMD distribution at various downstream stations at PW= 65psi and P jf= 25 psi for nozzle #100.

SMD RADIAL DISTRIBUTION

X:33 CNo X--53 CN

•4 -2 6RADIAL DISTANCE (CN)

psi

Figure 26 Radial SMD distribution at various downstream stations at P = 200psi and Pajr= 25 psi for nozzle #100.

-8

SHD RADIAL DISTRIBUTION

-4 -2RADIAL DISTANCE (CN)

4

psi, psi

Figure 27 Radial SMD distribution at various downstream stations at P = 200psi and P^.* 75psi for nozzle #100. w

SMD RADIAL DISTRIBUTION

4 <- - -1 *RADIAL DISTANCE

1 4 J 4 J

psi, pr.i

Figure 28 Radial SMD distribution at various downstream stations at Ppsi and P^ 45 psi for nozzle #100.

200

ORIGINAL PAGE S3OF POOR QUALITY

•«. '-.r''

•«̂ • • . i • *•.* •

• .4y?k~;-v. , •

Figure 29 Picture taken at 15 cm downstream from atomizerwith 59 psi water injection and 10 psi airinjection.

220

H•

0

*

BW

W

8

m NEASURED

D R-R CONFUTED

• NI CONFUTED

Iri1

a

S-

ea

1i

c."

El

S3

3

3̂

sI

m@

EJ

ea:'}

1 2 3 4 5 6 7 8 9 101112131415SIZE BAND

40

A30-

P

C28-W

D R-R

• NI

1 2 3 4 5 6 7 8 9 101112131415SIZE BAND

Figure 30 An example of a "good Fit" in predicting sizedistribution using the Rosin-Rammler mode.

229019 MEASURED

D R-R COMPUTED

• MI COMPUTED

1 2 3 4 5 6 7 8 9 101112131415SIZE BAND

48

a

0 5-

D R-R

• HI

I1 2 3 4 5 6 7 8 9 101112131415

SIZE BAND

Figure 31 At the edge of the sprays the Rosin-Rammlermode fails to predict large size drops.

20SND RADIAL DISTRIBUTION

18-

16-

o Pn=115psi

o Pn=U5psi

- 4 - 2 0 2 4RADIAL DISTANCE (CM)

Figure 32at d'StribUti°" H

3t Various differential pressures for nozzleat Pa.r- 65 psi and downstream location x=25 inches.

•

24

22

18

16

14

12

8

Pair=25psiPair=ll5psi

Pair=45psiPair=85psiPair=155psi

IB 28 38 40 SB 6 0 7 0DIFFERENTIAL PRESSURE (psU)

Figure 33 The variation of plane averaged SMD versusdifferential pressures (P - P . ) for NASA #2air-assist nozzle. w air

f -

CORRELATION RESULTS FOR NASA #2 NOZZLE USING SF7

(0c

D

CO

-.325p.<- m..94

0.55)

Figure 34 SMD correlation for NASA #2 nozzle usingSimmons' equation.